A human resource recruitment resource display terminal system

Abstract

The present application relates to the technical field of human resource management and artificial intelligence interaction, in particular to a human resource recruitment display terminal system.The system is credible and cannot be tampered with, uses blockchain smart contract and hash evidence to eliminate qualification fraud from the source, establishes decentralized trust, and at the same time, the system is accurate and in-depth in matching, innovatively introduces implicit behavior data such as line of sight and micro-expression, combines deep learning to dynamically correct matching weight, significantly improves the accuracy of human-post matching, provides immersive interaction, realizes the change from "seeing" to "experiencing", and evaluates the candidate's ability in all directions.The system is private and secure, uses local private sandbox and zero-knowledge proof to realize "data available but invisible", automatically destroys the original sensitive data after verification, at the same time, the system continuously evolves itself, based on the federal learning mechanism, realizes the collaborative optimization of the whole network model without leaking user data, and the system becomes more intelligent with use.

Description

Technical Field

[0001] This invention relates to the field of human resource management and artificial intelligence interaction technology, and more specifically, to a human resource recruitment resource display terminal system that integrates multimodal perception, blockchain trusted verification, multi-dimensional dynamic profile construction, and immersive interactive experience. Background Technology

[0002] With increasingly fierce competition in the talent market, traditional recruitment models have revealed many drawbacks in terms of efficiency, authenticity, and user experience: Verifying the authenticity of information is difficult: job seekers frequently forge their academic qualifications and certificates. Traditional manual verification or simple online queries are inefficient and difficult to prevent tampering. Companies lack an efficient and tamper-proof trust mechanism.

[0003] The current system relies on a single dimension for matching job candidates with resumes, neglecting implicit behavioral data of job seekers during the browsing process (such as eye contact, micro-expression reactions, and interaction frequency). This results in an inability to accurately uncover users' true intentions and potential traits, leading to low matching accuracy.

[0004] Lack of interactive experience: Traditional graphic or video presentations lack immersion, making it difficult for job seekers to intuitively experience the work environment, and for companies to conduct timely and contextual assessments of job seekers' actual skills (such as coding ability, adaptability and communication) through conventional means.

[0005] High privacy and security risks: The recruitment process involves a large amount of sensitive personal information (biometrics, ID number, detailed resume), and centralized storage makes it an easy target for attacks. In addition, it lacks a "data minimization" verification mechanism, and the consequences of leakage are serious.

[0006] The process is rigid and lacks evolution: the existing recruitment process is fixed and cannot be dynamically adjusted based on real-time feedback; the algorithm model relies on historical concentrated data for training, making it difficult to continuously optimize with new data while protecting privacy.

[0007] Therefore, there is an urgent need for an intelligent recruitment terminal system that can integrate trusted verification, multi-dimensional behavior perception, intelligent dynamic matching, immersive interaction, and strict privacy protection. Summary of the Invention

[0008] The purpose of this invention is to provide a human resource recruitment resource display terminal system to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A human resource recruitment resource display terminal system, the system includes: The main body of the intelligent interactive terminal integrates a multimodal sensor array and a holographic display unit to capture users' biometric data and behavioral interaction trajectories in real time, and present immersive recruitment resource content; The trusted data access module is configured to connect to an external distributed recruitment ecosystem network, obtain electronic resume data of job applicants and job requirement data of enterprises through a preset blockchain node interface, and use smart contracts to perform on-chain authenticity verification and hash storage of key qualification information. The multi-dimensional dynamic profile engine is communicatively connected to the trusted data access module and the main body of the intelligent interactive terminal, respectively, and is used to integrate explicit resume data and implicit behavioral data to construct a real-time dynamic profile of job seekers. The implicit behavioral data is vector data generated based on gaze lingering heatmaps, micro-expression emotional values ​​and body interaction frequencies collected by a multi-modal sensor array. The adaptive matching decision center has a built-in deep neural network matching model, which is used to receive the multi-dimensional dynamic profile, perform bidirectional matching calculations in combination with the enterprise's job requirements, dynamically adjust the matching strategy based on real-time feedback, and output hierarchical job recommendation results. An immersive resource display unit is used to automatically generate and render a personalized display interface that includes AR real-scene office roaming, job skill map visualization and virtual interviewer interaction based on the hierarchical job recommendation results. The end-to-end privacy sandbox module is built in the local isolated operating environment of the main body of the intelligent interactive terminal. It is used to de-identify user biometrics and sensitive resume data, and to verify specific qualifications with enterprises based on zero-knowledge proof technology. After verification, the original sensitive data is automatically destroyed.

[0010] In some embodiments, the main body of the intelligent interactive terminal is also equipped with an end-to-cloud collaborative relay interface: Supports establishing near-field communication connections with job seekers' personal mobile devices; When the system detects that a job seeker has finished interacting with the terminal and left the sensing area, it automatically encrypts and synchronizes the current browsing progress, matching analysis report, and interview appointment information to the job seeker's personal mobile terminal, and generates an electronic receipt with a timestamp.

[0011] In some embodiments, the trusted data access module further includes a cross-chain interoperability interface: Supports integration with educational authorities' academic qualification chains, industry association certificate chains, and former employer alliance chains; When a verification request for a diploma or certificate uploaded by an applicant is detected, the smart contract is automatically triggered to call the corresponding consortium blockchain node and return a hash value certificate containing the digital signature of the verification result, ensuring that the certificate verification process is tamper-proof and traceable.

[0012] In some embodiments, the specific process by which the multidimensional dynamic profiling engine constructs implicit behavioral data includes: By using eye-tracking sensors to record the duration of user gaze and pupil dilation rate when browsing different job detail cards, an interest heat vector is generated. Facial expression recognition algorithms are used to capture micro-expression fluctuations of users while viewing information, including salary, work location, and welfare policies, and to calculate sentiment tendency coefficients. The interest heat vector and the sentiment tendency coefficient are used as implicit weighting factors and superimposed on the initial matching model of the explicit resume data to adjust the ranking priority of job recommendations in real time.

[0013] In some embodiments, the immersive resource display unit supports content rendering based on generative artificial intelligence: For technical R&D positions, the system automatically calls the company's anonymized code library or project case library to generate an interactive code sandbox environment on the terminal screen for job seekers to take skills tests on-site. For marketing or service positions, AIGC technology is used to generate virtual customer scenarios and dialogue scripts in real time, simulate real interview questions and answers through voice interaction, and assess job seekers' adaptability and communication skills in real time.

[0014] In some embodiments, the system further includes a federated learning-based model evolution mechanism: Each distributed intelligent interactive terminal entity trains local matching model parameters locally using the current user's interaction data; Only the encrypted model parameter gradients are uploaded to the cloud central server for aggregation and updates, while the original user data is kept in a local privacy sandbox, enabling continuous algorithm optimization where data is available but not visible.

[0015] In some embodiments, the system further includes a visual process configuration module: Provides a graphical interface for enterprise human resource managers to define recruitment screening funnels; It allows binding automated trigger conditions at each stage of the screening funnel, including automatic termination upon diploma verification failure, automatic transfer to the talent pool when the implicit interest score is below a threshold, or automatic initiation of a video interview invitation when the matching degree is above a preset value.

[0016] This invention also provides a method for demonstrating the system, the method being: S1. Wake-up and Sensing: When the terminal is in sleep monitoring mode, it senses the user's approach through the ToF depth sensor, automatically wakes up and loads the enterprise brand holographic projection; S2. Identity Verification and Sandbox Creation: Collect user biometrics through multimodal sensors, combine with blockchain interface to quickly verify user identity and key qualifications, and create a temporary privacy sandbox locally; S3. Implicit Data Collection: During the user's browsing process, capture eye movements, micro-expressions, and body movements in real time, and dynamically update the user's implicit preference vector; S4. Dynamic matching calculation: Input explicit resume data and implicit preference vectors into the deep matching model to calculate the job-person matching degree in real time and generate a personalized recommendation list; S5. Immersive Interaction: Based on the recommendation results, dynamically render AR office scenes or launch AI mock interview sessions to collect user interaction feedback; S6. Verification and Evidence Storage: After the user confirms their intention, they send qualification verification credentials to the enterprise through zero-knowledge proof technology, complete the interview appointment, and store key interaction data on the blockchain. S7. Cleanup and Synchronization: When the session ends, the original biometric data in the local sandbox is destroyed, only the desensitized behavior analysis logs are retained for model iteration, and the results are synchronized to the user's mobile device.

[0017] In some embodiments, in step S4, if it is detected that a user's implicit preference vector for a certain position exceeds a preset threshold three times consecutively, the system will automatically trigger a priority recommendation mechanism, set the online status of the company's HR for that position to be able to chat directly, and prioritize the allocation of a video interview channel.

[0018] Beneficial effects: (1) This system is trustworthy and tamper-proof. It uses blockchain smart contracts and hash storage to eliminate qualification fraud from the source and establish decentralized trust. At the same time, this system provides accurate and in-depth matching. It innovatively introduces implicit behavioral data such as eye contact and micro-expressions, and combines deep learning to dynamically correct matching weights, which significantly improves the accuracy of matching people with jobs. It also provides an immersive interactive experience. Through AR roaming, AI code sandbox and virtual interviews, it realizes the transformation from "seeing" to "experiencing" and comprehensively evaluates the candidate's ability.

[0019] (2) This system is privacy and security. It adopts a local privacy sandbox and zero-knowledge proof to achieve "data is usable but not visible". After verification, the original sensitive data is automatically destroyed, which meets the highest security standards. At the same time, this system continues to evolve itself. Based on the federated learning mechanism, it achieves collaborative optimization of the whole network model without leaking user data. The more the system is used, the smarter it becomes. (3) The system has a flexible and controllable process, supports graphical configuration of recruitment funnel and automated triggering conditions, and can adapt to the personalized recruitment needs of different companies. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall system architecture provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the workflow of the multi-dimensional dynamic portrait engine in this embodiment of the invention. Figure 3 This is a flowchart illustrating the blockchain-based qualification verification and evidence storage process in an embodiment of the present invention. Figure 4 This is a flowchart illustrating the steps of the recruitment resource display method in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the interaction principle between the privacy sandbox and zero-knowledge proof in an embodiment of the present invention. Figure 6 This is a schematic diagram of the operating interface of the immersive resource display unit in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are intended to fully explain the technical implementation of the present invention, but are not intended to limit its scope of protection.

[0022] Example 1 System core architecture and hardware deployment.

[0023] This embodiment describes the system's basic hardware architecture and core module deployment.

[0024] like Figure 1 As shown, the main body of the system is an intelligent interactive terminal deployed in office building lobbies, university employment centers, or job fairs.

[0025] Hardware configuration: Multimodal sensor array: integrates a 120Hz infrared eye tracker (records gaze coordinates and pupil changes), a high-resolution wide-angle camera (runs a lightweight CNN model to recognize 7 basic emotions and 12 micro-expressions), and a ToF depth sensor (builds 3D point cloud to recognize gestures and distance).

[0026] Holographic display unit: It adopts a transparent OLED screen, supports light field reconstruction, and provides AR stereoscopic display with a 6DoF viewing angle.

[0027] Edge computing nodes: Built-in NPU (computing power ≥15 TOPS) and hardware-level trusted execution environment (TEE) to ensure that biological data is processed only in a local secure area.

[0028] Module collaboration: The trusted data access module connects to the education blockchain, industry association blockchain, and enterprise alliance blockchain through preset interfaces. When a user uploads a diploma, the module automatically calls a smart contract to compare the hash value on the blockchain and returns a signature credential.

[0029] The multi-dimensional dynamic profiler engine receives sensor data streams in real time. For example, when a user browses the "algorithm engineer" job posting, eye tracking records a dwell time of 45 seconds (threshold 10 seconds), a pupil dilation rate of 15%, and micro-expressions showing "excitement." The engine transforms this implicit data into a weight vector and adds it to the explicit resume matching score.

[0030] The end-to-end privacy sandbox automatically creates an isolated container at the start of a session. All original biometric images and scanned ID cards are decrypted and processed only within the sandbox, and only zero-knowledge proof (ZKP) credentials are output externally. After the session ends (the user leaves the sensing area), the sandbox triggers a secure erase command to overwrite the memory data.

[0031] Example 2 Detailed logic of multi-dimensional dynamic profiling and adaptive matching.

[0032] like Figure 2 The embodiment shown here details the specific implementation of the portrait construction and matching algorithm.

[0033] 1. Implicit data vectorization: Define a hidden vector: V imp (t)=[w1⋅T gaze w2⋅E micro w3⋅G gesture ].

[0034] T gaze Normalized value of the duration of gaze in the "Salary and Benefits" area.

[0035] E micro : Micro-expression emotion coefficient when viewing "welfare policies" (+1 for extreme satisfaction, -1 for disgust).

[0036] G gesture : Frequency of gesture interaction (e.g., the number of times "view details" is actively clicked).

[0037] 2. Dynamic weighted fusion: Initial matching score S init Calculated based on explicit resume data (educational background, skills keywords).

[0038] Real-time correction of S real =S init +β⋅LSTM(V imp The LSTM network is used to capture the time-series features of behavior (such as the process of a user going from hesitation to determination).

[0039] Priority recommendation mechanism: If a certain position's V impIf the threshold is exceeded three times consecutively (e.g., the user repeatedly views the page and displays a positive expression), the system will automatically trigger a priority recommendation mechanism. This mechanism will set the HR's online status for that position to "available for direct chat" and prioritize the allocation of a video interview channel, without requiring the user to manually schedule an appointment.

[0040] 3. Adaptive adjustment: If a user frequently skips a certain type of job in the recommendation list (negative frequency of physical interaction), the matching model automatically reduces the weight of that type of job and refreshes the recommendation list in the next second, reflecting the characteristic of "the more you use it, the better it understands you".

[0041] Example 3 like Figure 3 The process shown ensures the authenticity of key information such as academic qualifications and certificates, while protecting the original documents from being leaked.

[0042] Step S301: Local file hashing Job seekers upload original electronic files (such as PDFs or images) of their diplomas or qualifications to the terminal touchscreen.

[0043] The key to this system is that files are not uploaded to any remote server. The privacy sandbox inside the terminal immediately calls the SHA-256 algorithm to perform a hash operation on the local file, generating a unique digital fingerprint H. user = Hash(File original ).

[0044] Step S302: On-chain query request The terminal sends a query request to the consortium blockchain node through the trusted data access module.

[0045] The request packet contains only H user And a file type identifier (such as "Bachelor's Degree"), but does not contain the original content of the file.

[0046] Step S303: Smart Contract Comparison Smart contracts on the blockchain receive H user Then, the official evidence hash value stored on the chain is automatically retrieved (this value is pre-stored by the issuing institution, such as a university or human resources and social security bureau, when it is uploaded to the chain).

[0047] Contract execution comparison logic: Determine whether Huser=Hchain is true.

[0048] Step S304: Branch processing If inconsistent (No): The smart contract returns a "verification failed" status code. The terminal screen displays a red warning: "Materials suspected of being tampered with or not registered on the chain," the process terminates, and the hash log of this abnormal behavior is recorded on the chain to prevent malicious brute-force attacks; If they match (Yes): The smart contract returns a "verification passed" signature and timestamp; Sub-step S304-1 (Evidence Storage): The terminal packages the time, job ID, and signature of the verification result into a "trusted interaction record" and writes it into the blockchain to form an unalterable audit trail. Sub-step S304-2 (credential generation): The terminal generates a temporary, time-limited zero-knowledge proof token (ZKP token). This token only proves that "the user possesses a genuine and valid certificate of a certain type," without revealing the specific certificate number or name (unless authorized by the user). Step S305: Document Delivery The system sends the ZKP Token to the enterprise's recruitment system. The enterprise only needs to verify the token's signature to confirm the authenticity of the qualifications. The system cannot obtain the user's original diploma documents throughout the process, thus achieving "verification equals authorization, and authorization is not disclosed."

[0049] Example 4 General process for displaying recruitment resources.

[0050] like Figure 4 As shown, the full lifecycle control logic from the user approaching the terminal to the completion of delivery is described in detail.

[0051] Step S401: Terminal wake-up and multimodal perception When the ToF sensor detects a human body approaching (distance <1.5 meters), the terminal wakes up from sleep mode; The screen lights up the welcome screen and simultaneously activates the eye-tracking and facial recognition modules, entering a "silent observation period" to initially determine the user's age group and gaze focus. Step S402: Identity Verification and Privacy Sandbox Construction After a user selects "Start Job Search", the system dynamically creates an isolated privacy sandbox area in memory; Users initialize their identity via NFC reading of their ID card or facial biometrics. All collected biometric images (facial photos, fingerprint data) are immediately encrypted and stored in a sandbox. The key is held by the user's handheld device or a temporary password; the system server does not have the key. Step S403: Implicit Behavior Data Collection (Recurring Process) During the user's browsing process, the system enters a high-frequency data acquisition cycle: Record the duration of eye contact with keywords such as "salary", "location", and "technology stack"; Capture micro-expressions (surprise, indifference) of users when they see specific benefits (such as "flexible work arrangements"). Count the frequency of gesture interactions (number of times you actively click or zoom in to view); These data are converted into vector streams in real time and input into the multidimensional dynamic portrait engine; Step S404: Dynamic matching calculation and threshold determination The engine calculates the matching score S between the currently viewed job posting and the user profile in real time. The judgment logic is as follows: If S <Threshold low (Low threshold): If the system determines that there is insufficient interest, it will automatically reduce the weight of this type of job in the background and reduce such content in the next screen of recommendations; If S>Threshold high (High threshold): The system determines that the user has "high intent" and triggers the priority recommendation mechanism; Step S405: Immersive Interactive Rendering For highly relevant positions, the screen switches to immersive mode: The left side presents an AR office environment walkthrough (users can control the viewpoint with gestures); An AI virtual interviewer pops up on the right, initiating a conversation, such as, "We've detected that you're very interested in the R&D atmosphere. Would you like to simulate our code review process?" If the user agrees, the system will enter the code sandbox or scenario simulation test. Step S406: Verification, Evidence Storage and Confirmation of Intent If the user decides to deliver, the system triggers the blockchain verification process described in Example 3; After successful verification, the user clicks "Confirm Submission". The system generates an encrypted data packet containing "resume hash + behavioral score + verification token" and sends it to the enterprise. The act of submitting the application itself is also recorded on the blockchain as "non-repudiable" evidence of the job application; Step S407: Data Cleaning and Synchronization After the user leaves the sensing area (ToF detects no one) or clicks "End Session": Destruction: All original biological images, temporary decryption keys, and intermediate calculation variables within the privacy sandbox are immediately overwritten and cleared to zero; Synchronization: Only the desensitized structured report (excluding the original biological image) will be pushed to the user's mobile app via an encrypted channel; The terminal returns to sleep mode, waiting for the next user.

[0052] Example 5 The deep interaction principle between privacy sandboxes and zero-knowledge proofs.

[0053] like Figure 5 As shown, this demonstrates how data can be transformed between "available" and "invisible".

[0054] 1. Regional division and data isolation like Figure 5 As shown above, the user side includes three core databases: biometric database (face / iris), original sensitive resumes (including ID number and home address), and original diploma documents; These three databases are enclosed in a "local privacy sandbox" (the dotted shield box in the image). The sandbox is a hardware-level isolated environment (TEE), where external operating systems, network processes, and even enterprise administrators have no right to directly read the internal data.

[0055] 2. Challenge-Response Mechanism Company Question: For example Figure 5 As shown below, the company's recruitment system sends verification requests, such as: "Does this candidate have a master's degree or above?" or "Is this candidate over 25 years old?" Note that the company does not request "Please send the original graduation certificate" or "Please send a photo of your ID card".

[0056] Black-box computation: Request the zero-knowledge proof engine (black cube in the image) to be passed into the sandbox.

[0057] The engine reads the original diploma and ID card data inside the sandbox.

[0058] Execution logic judgment: IsMaster(Degree)==True and (CurrentYear−BirthYear)>25.

[0059] If the conditions are met, the engine uses the zk-SNARKs algorithm to generate a short mathematical proof string (Proof). During this process, the original data never leaves the sandbox, nor is it exposed in plaintext form to untrusted processes in memory.

[0060] 3. Results Output and Validation like Figure 5 As indicated by the middle arrow, the sandbox only outputs verification credentials.

[0061] Once the enterprise receives the voucher, it uses the publicly available verification key to perform a verification. Successful verification means the "conditions are met".

[0062] Key effect: Companies get a definite "yes / no" answer, but are completely unable to deduce the candidate's specific university, date of birth, or ID number. Figure 5 The "prohibited eye icon" on the Chinese enterprise side vividly illustrates this state of "seeing the result but not the original text".

[0063] 4. Automatic destruction closed loop Once verification is complete, the temporary computation context within the sandbox immediately self-destructs. Even if a hacker breaks into the system at this point, they will only obtain a bunch of random noise and will be unable to reconstruct the user's original private data. This mechanism completely solves the security risks of "resumes flying everywhere and privacy exposed" in traditional recruitment.

[0064] Example 6 Immersive interaction and generative AI applications.

[0065] like Figure 6 The diagram illustrates the specific implementation of the immersive display unit.

[0066] Technical R&D role scenario: When a "Java Development Engineer" is matched, an interactive code sandbox is automatically generated on the screen. The sandbox connects to the company's anonymized code repository, requiring the job seeker to fix a bug or implement a specific function within 5 minutes.

[0067] The system compiles code in real time, and calculates the time taken, code standardization, and execution results, which serve as the core basis for skills assessment and are directly included in the matching score.

[0068] Marketing job scenario: AIGC technology is used to generate virtual customer scenarios in real time. The virtual person plays the role of a picky customer and asks unexpected questions (such as "Your product is 20% more expensive than the competition, why should I buy it?").

[0069] The system uses speech recognition and sentiment analysis to assess job seekers’ response logic, speech rate stability and emotional control, generating a “response capability radar chart”.

[0070] AR Real-world Roaming: The system reconstructs corporate offices using NeRF technology. Users can "walk" in the virtual space using gestures and click to view workstation details, cafeteria menus, and gym facilities. Based on user profiles (such as a focus on life balance), the system automatically highlights relevant areas.

[0071] Example 7 Evolution of edge-cloud collaboration and federated learning.

[0072] This embodiment describes the system's scalability and evolution mechanism.

[0073] 1. End-to-end cloud collaboration: Users bind their devices to their personal mobile phones via Bluetooth / NFC. When the ToF sensor detects that the user has left the terminal (distance > 2 meters), the system automatically encrypts and packages the current browsing progress, matching report, and interview appointment information, and sends them to the user's mobile app.

[0074] Generate timestamped electronic receipts to prevent future disputes and allow users to continue to complete their information or view interview feedback on their mobile phones.

[0075] 2. Federated Learning Model Optimization: Each terminal uses the de-identified behavioral data of the current user (such as "the correlation between gaze duration and final onboarding") to train the local model gradient ΔW.

[0076] The gradient is homomorphically encrypted and then uploaded to the cloud aggregation server to update the global model parameters before being distributed to all terminals.

[0077] The key to this system is that the original user data (faces, original resumes) never leaves the local sandbox, and only the encryption gradient is transmitted, so as to achieve continuous evolution of "data available but not visible".

[0078] Example 8 Visualized process configuration and automated management.

[0079] This example describes the configuration capabilities on the enterprise side.

[0080] Provides a web-based graphical process configurator, allowing HR professionals to drag and drop components to define the recruitment funnel: Node 1: Diploma Verification. Configuration Conditions: If blockchain verification fails, terminate the process directly and record the blacklist hash.

[0081] Node 2: Implicit Interest Filtering. Configuration condition: If the "Interest Popularity Vector" is below 0.4, it will automatically be transferred to the "Potential Talent Pool" and a follow-up email will be sent.

[0082] Node 3: High Match Rate Direct Access. Configuration Conditions: If the overall match score is >90 and the code test passes, a video interview invitation will be automatically triggered, and the HR director will be notified to intervene.

[0083] This module enables non-technical personnel to flexibly customize recruitment strategies to adapt to the differentiated needs of different positions.

[0084] Comparative Example 1 Traditional recruitment terminal system.

[0085] System composition: A standard touch screen all-in-one machine, equipped only with a camera (for simple photo taking) and a display screen.

[0086] Data processing: ①Verification method: Relies on manual uploading of scanned documents, which are then visually verified by the HR department or manually queried through the China Higher Education Student Information System (CHESICC). There is no blockchain evidence, which poses a risk of forgery slipping through the net.

[0087] ② Matching logic: Static matching is performed solely based on resume text keywords (such as "Bachelor's degree" and "Python"), ignoring user browsing behavior. Even if a user shows great interest in a particular job (long-term gaze), the system will not adjust the recommendation ranking.

[0088] ③ Interaction format: Only static text and image introductions and video promotional videos are displayed; there is no AR roaming, no code sandbox, and no AI mock interview.

[0089] ④ Privacy Protection: Original user ID cards and resumes are directly uploaded to a centralized server for storage, without sandbox isolation or zero-knowledge proofs, resulting in a high risk of data leakage. The server retains all original data after the session ends.

[0090] Test result comparison: Matching accuracy: The average accuracy rate is only 65% ​​(due to a large number of false positives), while the accuracy rate of the embodiments of the present invention reaches 92%.

[0091] Counterfeit detection rate: The counterfeit detection rate is 80% (relying on manual verification), while the rate of this invention is 100% (based on blockchain hash comparison).

[0092] Average user dwell time: The comparative ratio is 1.5 minutes, while the present invention is 8.5 minutes (immersive interactive attraction).

[0093] Privacy compliance: Compared to the proportion that does not meet the "minimum data collection" principle, this invention fully complies with the GDPR and the Personal Information Protection Act.

[0094] Comparative Example 2 A semi-intelligent system without implicit behavior analysis.

[0095] System composition: It has blockchain verification function, but lacks multimodal sensor array (no eye tracking, micro-expression recognition).

[0096] Differences: Profile building: Relying solely on explicit resume data cannot capture the true fluctuations in a user's interests. For example, a user's resume may not mention "interested in AI," but their pupils dilate and they linger for a long time when browsing AI-related jobs. The comparison system cannot recognize this signal and will still recommend traditional jobs based on resume keywords, resulting in missing out on suitable candidates.

[0097] Dynamic adjustment: The recommendation list is static and fixed, and cannot be adjusted based on real-time user feedback (such as frequent changes). Adjustments will be made (by skipping certain job categories).

[0098] Test result comparison: Potential talent discovery rate: The ratio is only 40%. This invention uses implicit data analysis to discover 65% of candidates with imperfect resumes but high potential.

[0099] Interview conversion rate: The average conversion rate is 15%, but this invention achieves a conversion rate of 35% due to its more accurate recommendations.

[0100] Comparative Example 3 A cloud-based processing system without a privacy sandbox.

[0101] System composition: It has multimodal perception and AI matching, but no local privacy sandbox. All biometric and resume data are uploaded to the cloud for processing in real time.

[0102] Differences: ① Security: There is a risk of data interception or leakage by internal personnel during data transmission and storage.

[0103] ②Verification mechanism: Using traditional database comparison, companies can view the original complete resume of users, which violates the principle of "minimum necessary".

[0104] ③ Compliance: It cannot meet the increasingly stringent requirements for local data storage and the "right to be forgotten" (it is difficult to completely delete data in distributed backups).

[0105] Test result comparison: User trust rating: The comparison ratio is only 3.5 out of 5, and many high-end candidates refuse to use it; this invention scores 4.8 out of 5, and is highly trusted due to its zero-knowledge proof and automatic destruction mechanism.

[0106] Data leakage simulation attack: The comparative example leaks 100% of the original data under attack; the system of this invention only leaks meaningless encryption gradients or empty data that has been destroyed.

[0107] As can be seen from the comparison of the above embodiments and comparative examples, this invention significantly solves the pain points of traditional recruitment technologies in terms of authenticity, accuracy, user experience, and security through the deep integration of multimodal implicit perception, blockchain trusted storage, zero-knowledge proof privacy protection, and immersive generative interaction. The system not only achieves efficient job matching but also builds a trustworthy recruitment ecosystem, possessing extremely high practical value and promising prospects for widespread adoption.

[0108] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should be included in the scope of the present invention.

Claims

1. A human resource recruitment resource display terminal system, characterized in that, The system includes: The main body of the intelligent interactive terminal integrates a multimodal sensor array and a holographic display unit to capture users' biometric data and behavioral interaction trajectories in real time, and present immersive recruitment resource content; The trusted data access module is configured to connect to an external distributed recruitment ecosystem network, obtain electronic resume data of job applicants and job requirement data of enterprises through a preset blockchain node interface, and use smart contracts to perform on-chain authenticity verification and hash storage of key qualification information. The multi-dimensional dynamic profile engine is communicatively connected to the trusted data access module and the main body of the intelligent interactive terminal, respectively, and is used to integrate explicit resume data and implicit behavioral data to construct a real-time dynamic profile of job seekers. The implicit behavioral data is vector data generated based on gaze lingering heatmaps, micro-expression emotional values ​​and body interaction frequencies collected by a multi-modal sensor array. The adaptive matching decision center has a built-in deep neural network matching model, which is used to receive the multi-dimensional dynamic profile, perform bidirectional matching calculations in combination with the enterprise's job requirements, dynamically adjust the matching strategy based on real-time feedback, and output hierarchical job recommendation results. An immersive resource display unit is used to automatically generate and render a personalized display interface that includes AR real-scene office roaming, job skill map visualization and virtual interviewer interaction based on the hierarchical job recommendation results. The end-to-end privacy sandbox module is built in the local isolated operating environment of the main body of the intelligent interactive terminal. It is used to de-identify user biometrics and sensitive resume data, and to verify specific qualifications with enterprises based on zero-knowledge proof technology. After verification, the original sensitive data is automatically destroyed.

2. The human resource recruitment display terminal system according to claim 1, characterized in that, The main body of the intelligent interactive terminal is also equipped with an end-to-cloud collaborative relay interface: Supports establishing near-field communication connections with job seekers' personal mobile devices; When the system detects that a job seeker has finished interacting with the terminal and left the sensing area, it automatically encrypts and synchronizes the current browsing progress, matching analysis report, and interview appointment information to the job seeker's personal mobile terminal, and generates an electronic receipt with a timestamp.

3. The human resource recruitment display terminal system according to claim 1, characterized in that, The trusted data access module also includes a cross-chain interoperability interface: Supports integration with educational authorities' academic qualification chains, industry association certificate chains, and former employer alliance chains; When a verification request for a diploma or certificate uploaded by an applicant is detected, the smart contract is automatically triggered to call the corresponding consortium blockchain node and return a hash value certificate containing the digital signature of the verification result, ensuring that the certificate verification process is tamper-proof and traceable.

4. The human resource recruitment display terminal system according to claim 1, characterized in that, The specific process by which the multi-dimensional dynamic profiling engine constructs implicit behavioral data includes: By using eye-tracking sensors to record the duration of user gaze and pupil dilation rate when browsing different job detail cards, an interest heat vector is generated. Facial expression recognition algorithms are used to capture micro-expression fluctuations of users while viewing information, including salary, work location, and welfare policies, and to calculate sentiment tendency coefficients. The interest heat vector and the sentiment tendency coefficient are used as implicit weighting factors and superimposed on the initial matching model of the explicit resume data to adjust the ranking priority of job recommendations in real time.

5. The human resource recruitment display terminal system according to claim 1, characterized in that, The immersive resource display unit supports content rendering based on generative artificial intelligence: For technical R&D positions, the system automatically calls the company's anonymized code library or project case library to generate an interactive code sandbox environment on the terminal screen for job seekers to take skills tests on-site. For marketing or service positions, AIGC technology is used to generate virtual customer scenarios and dialogue scripts in real time, simulate real interview questions and answers through voice interaction, and assess job seekers' adaptability and communication skills in real time.

6. The human resource recruitment display terminal system according to claim 1, characterized in that, The system also includes a model evolution mechanism based on federated learning: Each distributed intelligent interactive terminal entity trains local matching model parameters locally using the current user's interaction data; Only the encrypted model parameter gradients are uploaded to the cloud central server for aggregation and updates, while the original user data is kept in a local privacy sandbox, enabling continuous algorithm optimization where data is available but not visible.

7. The human resource recruitment display terminal system according to claim 1, characterized in that, The system also includes a visual process configuration module: Provides a graphical interface for enterprise human resource managers to define recruitment screening funnels; It allows binding automated trigger conditions at each stage of the screening funnel, including automatic termination upon diploma verification failure, automatic transfer to the talent pool when the implicit interest score is below a threshold, or automatic initiation of a video interview invitation when the matching degree is above a preset value.

8. The human resource recruitment display terminal system according to any one of claims 1 to 7, characterized in that, The system is demonstrated as follows: S1. Wake-up and Sensing: When the terminal is in sleep monitoring mode, it senses the user's approach through the ToF depth sensor, automatically wakes up and loads the enterprise brand holographic projection; S2. Identity Verification and Sandbox Creation: Collect user biometrics through multimodal sensors, combine with blockchain interface to quickly verify user identity and key qualifications, and create a temporary privacy sandbox locally; S3. Implicit Data Collection: During the user's browsing process, capture eye movements, micro-expressions, and body movements in real time, and dynamically update the user's implicit preference vector; S4. Dynamic matching calculation: Input explicit resume data and implicit preference vectors into the deep matching model to calculate the job-person matching degree in real time and generate a personalized recommendation list; S5. Immersive Interaction: Based on the recommendation results, dynamically render AR office scenes or launch AI mock interview sessions to collect user interaction feedback; S6. Verification and Evidence Storage: After the user confirms their intention, they send qualification verification credentials to the enterprise through zero-knowledge proof technology, complete the interview appointment, and store key interaction data on the blockchain. S7. Cleanup and Synchronization: When the session ends, the original biometric data in the local sandbox is destroyed, only the desensitized behavior analysis logs are retained for model iteration, and the results are synchronized to the user's mobile device.

9. The human resource recruitment display terminal system according to claim 8, characterized in that, In step S4, if the system detects that a user's implicit preference vector for a certain position exceeds a preset threshold three times consecutively, it will automatically trigger a priority recommendation mechanism, set the online status of the company's HR for that position to be available for direct chat, and prioritize the allocation of a video interview channel.

Images